Page History

The new Juno cluster is now available to the test users only. Please, create a request if you want to participate. It is currently composed of 216 CPUs. The login node name is juno. It's a CentOS 7.3 host. The latest CentOS 7.3 Linux operating system is on nodes jx01-jx10  and the Centos 6.9 Linux operating system is on node ju14 . All cluster nodes have access to Isilon solisi file system. The CentOS 7.3 nodes have access to the GPFS storage named juno.

Differences in LSF configuration

...

between Juno

...

and Luna

1. GPFS and OS We reserve ~12GB of RAM per host on Juno CentOS7 hosts. Usable memory for LSF jobs per node type: jx* have 240GB and ju* have 250GBMemory (GB) is for the operating system and GPFS on Juno CentOS 7 hosts.
2. On each jx## node (CentOS 7, GPFS installed), 240GB of RAM is available for LSF jobs.
3. On each ju## node (CentOS 6, no GPFS), 250GB of RAM is available for LSF jobs.
4. When specifying RAM for LSF jobs, specify GB of RAM per slot/task on Juno. Memory(GB) is , or per job on Luna.
5. All jobs should must have -W (maximum Walltime) specified on Juno. LSF enforces it. Please , don't do not use -We on Juno.
6. To check job information for jobs which are DONE or status EXIT jobs , use "bhist `bhist -l JobIDJobID`" or "bhist `bhist -n 0 -l JobIDJobID`". Also "bacct" can be used`bacct` is also available. " bjobs `bjobs -l JobIDJobID`" should be used only for RUNNING or PEND only shows `RUNNING` and `PEND` jobs.
7. There is no `/swap swap` on CentOS7 CentOS 7 nodes. Memory usage is enforced by cgroups on hosts and job won't be able to use /swap. Job , so job never use `/swap`. A job will be terminated on kernel level if job memory usage exceeds requested amount on its LSF specification.
8. There is no "iounits" `iounits` resource on Juno.

Queues

The Juno cluster uses LSF (Load Sharing Facility) 10.1  FP6 from IBM to schedule jobs. The default LSF queue, ‘general’, includes all Juno compute nodes.

...